Perceptions and Preferences for Meritocracy
1 Introducción
La desigualdad económica se ha vuelto un tema que genera creciente preocupación y malestar alrededor del mundo. Esto se ha expresado en una serie de protestas como la emblemática “occupy wall street” el año 2011, así como también en una serie de análisis críticos respecto del desarrollo del capitalismo y sus consecuencias (Piketty 2014). En este contexto, el estudio de las visiones, preferencias y percepciones respecto de la desigualdad han adquirido relevancia en las ciencias sociales, en temas como las preferencias redistributivas (???; ???) la legitimación de la desigualdad económica (???) y el funcionamiento de la meritocracia (Duru-Bellat and Tenret 2012; Mijs 2016; Reynolds and Xian 2014).
En general, la meritocracia se define como un sistema de distribución de recursos y recompensas basados en el mérito individual, que en su concepción original es una suma de talento y esfuerzo (Young 1962) y que pone en un lugar secundario interferencia de factores estructurales como la herencia o los contactos (Breen and Goldthorpe 1999; Saunders 1995; Yair 2007; Land 2006; Young 1994). Una serie de estudios han realizado críticas a la realización de este estándar moral de distribución, planteando que es una promesa incumplida dada la influencia preponderante de otros elementos más allá del mérito en el estatus individual (Arrow, Bowles, and Durlauf 2000; Goldthorpe 2003; Markovits 2019). Por otro lado, desde la psicología social y la sociología se han estudiado las características y consecuencias de las creencias en la meritocracia, en general basados en la hipótesis que mayor creencia en la meritocracia lleva a una mayor legitimación de las desigualdades (Hadjar 2008; Madeira et al. 2019).
Para poder dar cuenta de los niveles de creencia en la meritocracia los estudios a la fecha generalmente han utilizado algunos indicadores de encuestas ya existentes, y en el menor de los casos se han creado instrumentos ad-hoc. Sin embargo, y como mostraremos más adelante, las formas de medición de meritocracia varían extremadamente entre estudios. Muchas veces fenómenos similares se asocian a indicadores distintos, y también ocurre que fenómenos distintos son medidos con indicadores similares, todo lo cual hace dificulta la comparabilidad entre estudios y el poder avanzar en la comprensión del concepto de meritocracia.
Basados en el análisis crítico de las formas de medición de meritocracia a la fecha, el presente estudio propone un instrumento para medir y relacionar dos aspectos claves en el estudio de la meritocracia: percepciones y preferencias. Además, como un segundo eje de análisis considera la generación de indicadores respecto de aspectos meritocráticos y anti-meritocráticos, demostrando que no son los dos polos de un mismo continuo como muchos estudios anteriores parecen sugerir.
2 La medición de los aspectos subjetivos de la meritocracia
A continuación se presenta una revisión de una serie de investigaciones que se han abocado al estudio de la meritocracia y que para ello han hecho una propuesta de medición. El primer eje de análisis tiene que ver con el uso del concepto “creencias” para referir a distintos aspectos subjetivos relacionados con meritocracia. El segundo eje tiene que ver con el uso de indicadores sobre aspectos anti-meritocráticos como el polo opuesto de los meritocráticos.
2.1 La caja negra de las creencias en la meritocracia
Para ejemplificar la confusión del uso de creencias en meritocracia consideremos en primer lugar un artículo reciente de Mijs, donde se expresa la siguiente definición: “when I discuss meritocracy beliefs, I am referring to citizens’ belief in the importance of hard work relative to structural factors.” (Mijs 2019, pg.9). In the operationalization, this is associated with the following indicator: “how important you think it is for getting ahead in life: (a) hard work”, scored in a 1 to 5 likert scale. There are several assumptions behind this decision that are worth discussing:
Dimensionality: the item used by Mijs is part of an battery present in several international surveys usually called “get ahead”, which presents a series of indicators related to what people consider important to get ahead in life: hard work, education, ambition, wealthy family, right connections, religion, race and gender. For Mijs, other aspects as education, that could be associated to talent, are not meritocratic: “Hard work is arguably the most meritocratic part of Michael Young’s equation, ‘Merit = Intelligence + Effort’, for the simple fact that intelligence itself is conditioned by a nonmeritocratic factor: who your parents happen to be” (p.5). This is a very strong assumption: effort would not depend on parents influence, and talent is not meritocratic. A similar approach only considering effort as the main element of meritocracy is taken by (???). We suggest here is that questions such as whether talent is not meritocratic (or less meritocratic than effort) should be open to empirical scrutiny and not disregarded upfront based on scholars’ assumptions. Besides, the use of single indicators for measurement a construct is based on the assumption of a measurement without error.
Beliefs: the “get ahead” battery refers to “how important you think it is”, considered by Mijs as a belief in meritocracy. Nevertheless, another version of this same battery asks about “how important you think it should be”. Which one of both is the “belief”? If we consider belief as related to a normative content, actually the more close version is the “should be”, being the other one more close to the perception of the situation. Such distinction was already made by Duru-Bellat and Tenret (2012), who used the item “how important should the number of years spent in education and training be in deciding how much money people ought to earn?” for “desired” meritocracy, whereas for “perceived” meritocracy they use two items: “Would you say that in your country, people are rewarded for their efforts?” and “… people are rewarded for their skills?”. Is the belief in meritocracy a perception or a preference? A comprehensive measurement should be in line with what Duru-Bellat and Tenret (2012) did, in order to open up possibilities of analyzing whether perceptions and preferences are actually the same (i.e. correlation close to 1) or they are different aspects of the same phenomenon. As Son Hing et al. (2011) have pointed out, “People can believe that outcomes ought to be distributed on the basis of merit and yet vary in their perceptions of whether this is how society currently operates” (p. 435). Accordingly, we suggest that the use of the term belief is confusing, being more accurate to speak about meritocratic perceptions and preferences.
Non-meritocratic aspects: Mijs (2019) make reference to some non-meritocratic aspects as talent, letting this out of the measurement. A different approach was followed by Kunovich and Slomczynski (2007), who decide to include non-meritocratic elements in the measurement of meritocracy. Using the items’ battery listing a number of reasons about “How important should be in deciding pay…” (as Duru-Bellat and Tenret (2012) for desired meritocracy), he decided that reasons as education and responsibility are meritocratic and pointed 1 if considered essential, whereas reasons such as having a family and children were pointed 1 if they were considered “not important at all” (i.e. reverse coded). A similar approach was taken by Newman, Johnston, and Lown (2015), reverse-coding non-meritocratic items, similarly to what occurs with the “Preference for the Merit Principle Scale” (Davey et al. 1999). The assumption that meritocratic and non-meritocratic elements are the poles of the same continuum was analyzed by Reynolds and Xian (2014) using the same “get ahead” perceptions’ battery items. They consider education, ambition and hard work as meritocratic and other reasons such as wealthy family and right connections and non-mertitocratic. Nevertheless, despite making this distinction the author ends up substracting one dimension from the other, assuming that they are two poles of the same continuum as Kunovich and Slomczynski (2007) did.
Accounting for measurement error: finally, most of the studies in meritocracy so far have not incorporated the issue of measurement error (???), using single indicators and/or composite indexes for measuring meritocracy. This means assuming that the construct is measured perfectly by the indicators chosen, going as far as proposing that "we decided to rely on a summative index. In choosing this strategy of index construction, we argue that support for meritocracy is not a latent variable (Kunovich and Slomczynski 2007, 653–54). Some advances were done by Reynolds and Xian (2014) by doing a principal component analysis of meritocratic and non-meritocratic dimensions, but somewhat contradictorily they end up in a sum index despite proving multidimensionality.
2.2 An instrument proposal
Based on the previous limitations in the measurement of meritocracy presented in the previous section, in this paper we propose and test an instrument with the following characteristics:
- Multidimensional, incorporating previous distinctions between preferences and perceptions as well as between meritocratic and non-meritocratic aspects.
- Multiple indicators for each dimension, in order to account for measurement error in a confirmatory factor analysis context.
- Based on previous indicators as far as possible, for the sake of comparability between studies
- Brief, as to be used in a regular survey. In this point it differs for instance from the proposal of the “Preference for the Merit Principle Scale” (Davey et al. 1999), as they use 15 items just for one dimension (besides the problem of reverse-coding non-meritocratic items).
The proposed measurement model is based on two axis of analysis, as depicted in Table 1:
Table 1: Model of perceptions and preferences for meritocracy and non-meritocracy
| Perceptions | Preferences | |
|---|---|---|
| Meritocracy | ||
| Non-meritocracy |
The first axis distinguishes between different types of “beliefs”, using instead the terms perceptions and preferences for meritocracy (Duru-Bellat and Tenret 2012; Son Hing et al. 2011). Perceptions refers to the extent to which people see meritocracy working in their society, which in terms of measurement relates to the “reasons to get ahead” battery, whereas preferences refer to normative expectations that are usually linked to a “should” expression (e.g. whether hard work should be related to payment). The second axis consider the distinction between meritocratic and non-meritocratic dimensions (Reynolds and Xian 2014). This aspect has been usually treated as different ends of a same continuum in part of the previous research, an assumption that requieres empirical scrutiny. These non-meritocratic elements usually refer to the use of personal contacts or family advantages to get ahead in life.
Regarding the selection of indicators, most of them some adaptations of previous items that where streamlined in order to a better fit with the specific dimensions. For meritocratic indicators we use effort and talent as the main components of the traditional concept of merit as defined by Young (1962), whereas for non-meritocratic dimensions we use having rich parents and good contacts.
Hypotheses
H1. The perception of meritocracy is a latent variable based on indicators of the importance attributed to talent and the effort to get ahead in life.
H2. The non-meritocratic perception is a latent variable that derives from two indicators related to the agreement with the statement that people with contacts and rich parents manage to get ahead.
H3. Meritocratic preferences behave as the latent variable based on a normative value of effort and talent.
H4. Non-meritocratic preferences behave as a latent variable based on the normative value of the use of personal contacts and having rich parents.
3 Methodology
The study design consist in two main stages. The first stage was the questionnaire design, in which we collect a serie of background variables, including sex, age and educational level. On this regard, one of the main aspects of our design is that we tested three different application modalities, in order to observe how the display order of the indicators could affect the model fit. We decide to randomize the assignment of the subject to each application modality, and as a result we obtain three groups within the full sample. The first combination corresponds to the group that answered the questions that were sorted by the Perceptions/Preferences dimension. The second combination corresponds to the group in which the questions are displayed following a topic-sorted way, this means that the effort, talent, rich family and networks questions for perception and preference were displayed together. The last and third group corresponds to a completely randomized display order of the questions. The figure below describes the complete survey flow.
Figure: Study design
To contact the subjects, we followed an online programmed survey approach using Qualtrics, and the field work was conducted by Netquest, which is a company with wide experience applying online survey, and for the case of Chile, they have a consolidated pool of individuals that systematically respond online surveys. On this regard, the company have their own incentive system, in which the subjects accumulates specific currencies that they can exchange for a wide range of goods.
3.1 Data
The Netquest company (website) applied the online survey during December and January. The online survey distributed by the company was scheduled on the platform Qualtrics (website). The sample was selected from a non-probabilistic design in three large cities in Chile. The quota method based on age, sex and educational level was used. Quotas were generated based on the survey of the Public Studies Center (CEP, 2019), which has a high prestige in the country. A total sample of 2141 people was collected, excluding those who did not answer the questions on the scale and those who did not accept informed consent. The representativeness of the sample is not satisfactory because, although the proportion by age and sex is consistent with the estimates of the CEP survey, a bias towards high educational levels is evident (as detailed in Annex X). This limitation must be considered when making inferences about the population.
Anexo x
| Sample | CEP | |
|---|---|---|
| gender | ||
| Men | 49,82% | 50,52% |
| Women | 50.18% | 49,47% |
| Age | ||
| 18 - 24 | 18,55% | 18,17% |
| 25 - 34 | 18,86% | 17,48% |
| 35 - 44 | 19.09% | 19,98% |
| 45 - 54 | 17,96% | 19,23% |
| 55 - or more | 25,54% | 25.11% |
| Education | ||
| Primary or less | 2,93% | 15,88% |
| Hig school | 43,23% | 37,04% |
| Non university | 32,63% | 28,93% |
| university or more | 21,21% | 18,13% |
Based on the recommendations of Wilson, Strite and Loiacono (2017) it has been considered pertinent transparency that the order the items of the meritocracy scale in the survey was random for a third of the sample. With the objective of evaluating the effect of the order of the indicators, the respondents were randomly divided into three groups as explained in Diagram 1. The scale was presented to the first group in the order that appears in Table 2, in such a way that the items of the same factor are found together. The second group was presented with the scale with the items of the same factor separated from each other, in such a way that they were consulted at the same time for their perceptions and preferences of a particular topic (e.g. distribution according to effort). The third group was presented with the scale with the randomized items to assess the possible bias produced by the order in which the questions are presented ((Budd, 1987) [https://psycnet.apa.org/record/1988-31448-001]; Davis and Venkatesh 1996; Goodhue and Loiacono, 2002; Schriesheim et al., 1989).
!!! Diagrama de flujo de la encuesta, orden de la escala.
3.2 Variables
The proposed scale of perceptions and preferences about meritocracy consists of 8 indicators that are grouped into the 4 dimensions indicated. The questions are adaptations of the “get ahead” scale. the terms of of this scale were used (i.e talent, effort, contacts and wealthy parents) which are included in different investigations on meritocracy (Mijs 2019; Duru-Bellat and Tenret 2012; Reynolds and Xian 2014). Unlike this scale, this proposal adds the difference between perceptions and preferences. In addition, a comparative component is incorporated into the question that refers to differences between people with little or fed up with some attribute (e.g. effort, contacts).The eight items ordered according to dimensions are presented in Table 2. These 8 likert-type items have 5 response alternatives ranging from “I completely agree” (5) to “I completely disagree”(1). According to the recommendations of Xia and Yang (2018) for categorical data, we present below a table with the descriptive table and another with correlations between the items on the scale.
The field work was conducted during November and December of 2019 in Chile, obtaining a final sample of 2200 individuals, but considering that we use cases with complete information, the final analytical sample has 2141 observations.
| Variable | Mean | SD | Min | Max | A | B | C | D | E | F | G | H |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| A. Who the more they try they manage to get bigger rewards that those who striveless. | 3.20 | 1.38 | 1 | 5 | \(-\) | |||||||
| B. Who possess more talent they manage to obtain greater rewards than those who possess less talent. | 3.02 | 1.16 | 1 | 5 | 0.52*** | \(-\) | ||||||
| C. Who they have rich parents manage to get out ahead. | 3.66 | 1.36 | 1 | 5 | -0.06* | 0.07** | \(-\) | |||||
| D. Who they have good contacts they manage to get out ahead. | 3.79 | 1.24 | 1 | 5 | -0.02 | 0.05** | 0.73*** | \(-\) | ||||
| E. Who the more they try they should get greater rewards than those who they try less. | 3.89 | 1.25 | 1 | 5 | 0.40*** | 0.28*** | 0.30*** | 0.34*** | \(-\) | |||
| F. Who possess more talent they should get greater rewards than those who possess less talent. | 3.24 | 1.19 | 1 | 5 | 0.17*** | 0.31*** | 0.27*** | 0.26*** | 0.49*** | \(-\) | ||
| G. It’s fine that those who have rich parents get ahead | 2.69 | 1.18 | 1 | 5 | 0.15*** | 0.14*** | 0.03 | 0.05** | 0.07** | 0.15*** | \(-\) | |
| H. Is well that those who have good contacts get ahead. | 2.41 | 1.11 | 1 | 5 | 0.16*** | 0.15*** | -0.09** | -0.03 | -0.01 | 0.10*** | 0.61*** | \(-\) |
| Note: \(N\)= 2141; p<0.05=\(*\); p<0.01=\(**\); p<0.001=\(***\) |
| Statistic | N | Mean | St. Dev. | Min | Pctl(25) | Median | Pctl(75) | Max |
| A. pc.effort | 2141 | 2.00 | 0.82 | 1 | 1 | 2 | 3 | 3 |
| B. pc.talent | 2141 | 3.20 | 1.38 | 1 | 2 | 3 | 4 | 5 |
| C. pc.wpart | 2141 | 3.02 | 1.16 | 1 | 2 | 3 | 4 | 5 |
| D. pc.netw | 2141 | 3.66 | 1.36 | 1 | 3 | 4 | 5 | 5 |
| E. pf.effort | 2141 | 3.79 | 1.24 | 1 | 3 | 4 | 5 | 5 |
| F. pf.talent | 2141 | 3.89 | 1.25 | 1 | 3 | 4 | 5 | 5 |
| G. pf.wpart | 2141 | 3.24 | 1.19 | 1 | 2 | 3 | 4 | 5 |
| H. pf.netw | 2141 | 2.69 | 1.18 | 1 | 2 | 3 | 4 | 5 |
| pref_netw | 2141 | 2.41 | 1.11 | 1 | 2 | 2 | 3 | 5 |
CFA
MGCFA
(???) (???)
3.3 Methods
We tested theoretical model of the scale of perceptions and preferences about meritocracy using confirmatory factor analysis (CFA) with for latent factor. CFA was conducted using the lavaan package (version 0.6-3; Rosseel, 2020) with diagonally weighted least squares (DWLS) estimation due to the items’ ordinal level of measurement (Kline, 2016; Rosseel, 2020). As recommended by Brown (2008), we assessed model fit by jointly considering the comparative fit index and Tucker-Lewis Index (CFI and TLI; acceptable fit > 0.95), Root of the average squared residual approximation (RMSEA; acceptable fit < 0.08), Chi-square: (p-value; acceptable fit > 0.05, and Chi-square ratio:> 3).
4 Results
| Variables | 1 | 2 | 3 | 4 | 1 | 2 | 3 | 4 | 1 | 2 | 3 | 4 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| A. Who the more they try they manage to get bigger rewards that those who striveless. | 0.69 | 0.76 | 0.70 | |||||||||
| B. Who possess more talent they manage to obtain greater rewards than those who possess less talent. | 0.81 | 0.72 | 0.65 | |||||||||
| C. Who they have rich parents manage to get out ahead. | 0.85 | 0.84 | 0.81 | |||||||||
| D. Who they have good contacts they manage to get out ahead. | 0.94 | 0.81 | 0.89 | |||||||||
| E. Who the more they try they should get greater rewards than those who they try less. | 0.85 | 0.82 | 0.66 | |||||||||
| F. Who possess more talent they should get greater rewards than those who possess less talent. | 0.64 | 0.65 | 0.59 | |||||||||
| G. It’s fine that those who have rich parents get ahead | 0.55 | 1.04 | 0.78 | |||||||||
| H. Is well that those who have good contacts get ahead. | 1.26 | 0.52 | 0.77 | |||||||||
| \(\chi^2\text{(df)}\) | 42.28(14) | 107.57(14) | 63.34(14) | |||||||||
| \(\text{CFI}\) | 0.993 | 0.961 | 0.979 | |||||||||
| \(\text{RMSEA}\) | 0.053 | 0.097 | 0.070 | |||||||||
| \(N\) | 712 | 717 | 712 | |||||||||
| Note: ; p<0.05=\(*\); p<0.01=\(**\); p<0.001=\(***\). Standardised factor loadings using DWLS estimator. CFI = Comparative fit index (scaled), RMSEA = Root mean square error of approximation (scaled) |
Este estudio cuenta con una encuesta panel de 3 olas. En la primera de los olas, se incorporó la escala en distintos órdenes. El primer orden, pregunta primero por todas las percepciones y después por todas las preferencias, el segundo orden, pregunta seguidamente por percepciones y preferencias en torno a un mismo tema y, el tercer orden incorpora los ítems de manera aleatoria. En la segunda y tercera ola, se utilizó el orden
Para poder evaluar la estructura factorial de la escala propuesta, se realizará un análisis factorial confirmatorio, evaluando el ajuste del modelo propuesto a partir de los estadísticos CFA, TLI, RMSEA y χ2 ratio. Los criterios de ajuste utilizados fueron anteriormente reportados en un preregistro de la investigación en el sitio Open Science FrameWork. Basado en los criterios de Kline (2016) y de Brown (2008) esperamos tanto para la prueba CFA como TLI, valores por sobre .95, mientras que para la prueba RMSEA, se consideran aceptables valores bajo .08 e idóneos bajo .05. Por su parte, respecto al χ2, se esperan valores p no significativos como parámetro ideal y se considera aceptable un χ2 ratio menor a 3.
En una primera instancia se probó el ajuste del modelo con tres distintos órdenes de los ítems, los cuales fueron aplicados a un tercio de la muestra de la primera ola cada uno. El primer orden de los ítems corresponde al orden de de la tabla 2, en el que primero se pregunta a los encuestados por todas las percepciones y después por todas las preferencias, en el segundo orden se les pregunta seguidamente por percepciones y preferencias respecto a un mismo tema, por ejemplo, al papel del esfuerzo, y en el tercero, los ítems aparecen de manera aleatoria.
Table 3: Fit indicators
| Model order 1 | Model order 2 | Model order 3 | |
|---|---|---|---|
| n | 697 | 702 | 697 |
| CFI | 0.997 | 0.984 | 0.991 |
| TLI | 0.992 | 0.968 | 0.984 |
| RMSEA | 0.037 | 0.071 | 0.051 |
| χ2 | 28.03 | 64.156 | 39.090 |
| p | .000 | .000 | .000 |
| χ2/gl | 2,00 | 4,58 | 2,78 |
Todos los modelos independiente de los órdenes obtuvieron un ajuste moderado con CFI superiores a .95 y RMSEA inferiores a .8, por contraparte ni un modelo logró un chi-square no significativo, aunque tanto el modelo aleatorio como el primero obtuvieron un adecuado chi-square ratio menor a 3. El primero de los órdenes fue el que mejor ajuste obtuvo (CFI=0.998, TLI= 0.995,RMSEA=0.037, χ2(df=14)=28,03, p = 0.014), seguido por el orden aleatorio de los ítems (CFI=0.992, TLI=0.984,RMSEA=0.051, χ2(df=14)=39.09, p < 0.001). Por su parte la escala ordenada por temáticas genera un efecto framing en el cual la relación entre las percepciones y las preferencias parecen sobreestimadas, lo cual afecta el ajuste, siendo este deficitario, con un χ2 ratio mayor a tres y un RMSEA que se puede encontrar por sobre .8 según el intervalo de confianza correspondiente(CFI=0.984, TLI=0.968,RMSEA=.071, χ2(df=14)=64.156, p < 0.001).
A partir de los análisis de invarianza se puede decir que los distintos órdenes no son invariantes en términos configuracionales (???). El modelo que incorpora la diferenciación de los grupos en la estimación del modelo, este presenta un ajuste moderado similar al de los modelos por separados, aunque posee un ratio de chi-square deficiente por lo que se descarta la invarianza. Esta varianza configuracional se debe a la sobre estimación que se genera en el orden uno de las correlaciones entre los indicadores de un mismo factor los cuales aparecen juntos para el encuestado, mientras que en el modelo 2 se presentan separados, disminuyendo la correlación entre los ítem de un mismo factor.Podríamos decir que las correlaciones entre los ítems son sub o sobre-estimadas según el orden en el que aparezcan. Coherentemente el modelo aleatorio tiene un ajuste intermedio entre ambos. Considerando la ventaja del orden, respecto aleatorio despeja la posibilidad del efecto framing, es decir, de que el resultado del modelo se deba al orden de las preguntas, se ha desidido seguir utilizando el orden 3.
| Contrast Model 1 Factor | theoretical model 4 Factors | Model with M.I. | |
|---|---|---|---|
| n | 1769 | 1769 | 1769 |
| CFI | 0.595 | 0.988 | 0.994 |
| TLI | 0.433 | 0.976 | 0.985 |
| RMSEA | 0.226 | 0.047 | 0.036 |
| χ2 | 1830.839 | 68.661 | 40.250 |
| p | .000 | .000 | .000 |
| χ2/gl | 65.38 | 4.90 | 3.32 |
EL modelo teórico propuesto de cuatro factores ajustó, como se observa en la tabla 4, mejor que el modelo de contraste de 1 factor. El modelo teórico ajustó de manera relativamente adecuada, pues muestra indicadores óptimos para CFI= 0.987, TLI = 0.975 y RMSEA=.041, aunque posee indicadores deficientes para la prueba χ2(df=14)=67.6, p-value=.000. Para evaluar posibles mejoras de la escala, se analizaron las relaciones propuestas por los índices de modificación. Estos indican la existencia de dos cargas cruzadas no especificadas. Cuando se generó un modelo siguiendo estas recomendaciones, hubo una mejora considerable del modelo, aunque el nuevo modelo tampoco obtuvo un X2 ratio menor a 3 y obtuvo cargas factoriales muy bajas ʎ < 0.15, por lo tanto, siguiendo las recomendación de Brown (2006) de solo aceptar las propuesta de los índices de modificación cuando se posee teoria y evidencia sólida, se ha decidido no incorporar estos parámetros al modelo.
- agregar cuáles fueron las cargas sugeridas.
5 Validez del constructo
5.1 Validez convergente
indicar la construcción de los índices – hacer un anexo.
Esta escala cuenta con validez de constructo, ya que los distintos índices formados según factor correlacionan coherentemente con la escala “Get a head”, utilizada por el programa internacional de encuestas sociales (ISSP), y con la escala de valores igualitarios utilizada por la encuesta internacional de valores.
| Igualitarismo | Get a head | |
|---|---|---|
| Percepción Meritocrática |
-0.143*** | 0.228*** |
| Percepción No Meritocrática |
0.130*** | -0.210*** |
| Preferencia Meritocrática |
-0.055** | 0.001 |
| Preferencia No Meritocratica |
-0.191*** | 0.083*** |
Existe coherencia entre el índice “Get a Head” y el índice de preferencias y percepciones meritocráticas. Como era esperable, la relación es positiva entre este índice y la percepción de meritocracia de la escala propuesta (r=0.228). En la misma línea, el índice Get a head, correlaciona de manera negativa con la percepción de no meritocracia (c=-0.210;p=0.000). Además, las relaciones con las preferencias son considerablemente menores, e incluso no es significativa respecto a las preferencias meritocráticas. En suma, el “Get a head” que es comúnmente utilizado para medir percepciones, posee relación solo con las variables de percepción y no así con las de preferencias.
Por su parte, el índice de Igualitarismo que refleja cuán problemática consideran la desigualdad los entrevistados, posee también relaciones coherentes. En primer lugar, es necesario destacar que solo posee una relación muy pequeña con la variable de preferencias meritocráticas, lo cual es positivo pues dicha dimensión de la escala no busca reflejar igualitarismo sino de preferencias meritocráticas. En segundo lugar, el índice se relaciona de manera negativa con la preferencia no meritocrática, lo cual es coherente en tanto alguien que valore la igualdad debería tender a no estar de acuerdo con la desigualdad de oportunidades. En tercer lugar, considerar la desigualdad como un problema, desde el índice de igualitarismo, posee como requisito creer que existe desigualdad, al respecto, es coherente que quienes consideran que la desigualdad existente es problemática, perciban igualmente que el éxito no depende sólo del esfuerzo y el talento (c=-0.143;p=0.000) sino que está influido por aspectos no meritocráticos (c=0.130;p=).
6 Discusión
7 Conclusión
7.1 Fiabilidad
7.1.1 Correlación entre los mismos items segun distintas olas.
| Pensando en la sociedad chilena, ¿En qué medida se encuentra usted de acuerdo o en desacuerdo con cada una de las siguientes afirmaciones? |
Ola 1 - Ola 2 |
Ola 2 - Ola 3 |
Ola 1 - Ola 3 |
|---|---|---|---|
| Quienes más se esfuerzan logran obtener mayores recompensas que quienes se esfuerzan menos. |
0.33 | 0.38 | 0.30 |
| Quienes poseen más talento logran obtener mayores recompensas que quienes poseen menos talento. |
0.30 | 0.28 | 0.27 |
| Quienes tienen padres ricos logran salir adelante. | 0.33 | 0.36 | 0.38 |
| Quienes tienen buenos contactos logran salir adelante. | 0.30 | 0.30 | 0.31 |
| Quienes más se esfuerzan deberían obtener mayores recompensas que quienes se esfuerzan menos. |
0.27 | 0.32 | 0.22 |
| Quienes poseen más talento deberían obtener mayores recompensas que quienes poseen menos talento. |
0.31 | 0.39 | 0.31 |
| Está bien que quienes tienen padres ricos salganadelante. | 0.40 | 0.41 | 0.39 |
| Está bien que quienes tienen buenos contactos salgan adelante. | 0.39 | 0.41 | 0.42 |
Pese a que existen bajas correlaciones, hay evidencia para apoyar la hipótesis de estabilidad de la escala a lo largo de las mediciones. Las correlaciones entre los mismos ítems en las distintas olas son de un tamaño pequeño a moderado según los parámetros de (???). Estas bajas correlaciones indican inestabilidad en las respuestas de los encuestados, no obstante existe evidencia para señalar una invarianza fuerte de las escalas en las distintas mediciones. En primer lugar, podemos decir que no existe diferencia entre el ajuste de los tres nos indica que no existe una diferencia significativa entre los ajustes al modelo de las tres olas, lo cual nos señala la existencia de invarianza configuracional.
Df AIC BIC Chisq Chisq diff Df diff Pr(>Chisq)fit.conf 42 35745 36222 110.91 fit.deb 50 35738 36172 119.39 8.485 8 0.3876 fit.fuer 58 35732 36124 129.94 10.546 8 0.2288 fit.str 74 35772 36079 202.02 72.081 16 4.298e-09 ***
El análisis de invarianza longitudinal muestra resultados moderadamente buenos. hay equivalencia configuracional, métrica, pero no escalar escalar
Bibliografía
Arrow, Kenneth J., Samuel Bowles, and Steven N. Durlauf, eds. 2000. Meritocracy and Economic Inequality. Princeton, N.J: Princeton University Press.
Breen, Richard, and John H. Goldthorpe. 1999. “Class Inequality and Meritocracy: A Critique of Saunders and an Alternative Analysis1.” The British Journal of Sociology 50 (1): 1–27. https://doi.org/10.1111/j.1468-4446.1999.00001.x.
Davey, L. M., D. R. Bobocel, L. S. Son Hing, and M. P. Zanna. 1999. “Preference for the Merit Principle Scale: An Individual Difference Measure of Distributive Justice Preferences.” Social Justice Research 12 (3): 223–40.
Duru-Bellat, Marie, and Elise Tenret. 2012. “Who’s for Meritocracy? Individual and Contextual Variations in the Faith.” Comparative Education Review 56 (2): 223–47. https://doi.org/10.1086/661290.
Goldthorpe, John. 2003. “The Myth of Education-Based Meritocracy.” New Economy 10 (4): 234–39. https://doi.org/10.1046/j.1468-0041.2003.00324.x.
Hadjar, Andreas. 2008. Meritokratie Als Legitimationsprinzip. Wiesbaden: VS Verlag.
Kunovich, Sheri, and Kazimierz M. Slomczynski. 2007. “Systems of Distribution and a Sense of Equity: A Multilevel Analysis of Meritocratic Attitudes in Post-Industrial Societies.” European Sociological Review 23 (5): 649–63. https://doi.org/10.1093/esr/jcm026.
Land, Hilary. 2006. “We Sat down at the Table of Privilege and Complained About the Food 1.” The Political Quarterly 77 (s1): 45–60. https://doi.org/10.1111/j.1467-923X.2006.00780.x.
Madeira, Ana Filipa, Rui Costa-Lopes, John F. Dovidio, Gonçalo Freitas, and Mafalda F. Mascarenhas. 2019. “Primes and Consequences: A Systematic Review of Meritocracy in Intergroup Relations.” Frontiers in Psychology 10 (September). https://doi.org/10.3389/fpsyg.2019.02007.
Markovits, Daniel. 2019. The Meritocracy Trap: How America’s Foundational Myth Feeds Inequality, Dismantles the Middle Class, and Devours the Elite. New York: Penguin Press.
Mijs, Jonathan J. B. 2016. “The Unfulfillable Promise of Meritocracy: Three Lessons and Their Implications for Justice in Education.” Social Justice Research 29 (1): 14–34. https://doi.org/10.1007/s11211-014-0228-0.
Mijs, Jonathan J B. 2019. “The Paradox of Inequality: Income Inequality and Belief in Meritocracy Go Hand in Hand.” Socio-Economic Review, January. https://doi.org/10.1093/ser/mwy051.
Newman, Benjamin J., Christopher D. Johnston, and Patrick L. Lown. 2015. “False Consciousness or Class Awareness? Local Income Inequality, Personal Economic Position, and Belief in American Meritocracy.” American Journal of Political Science 59 (2): 326–40. https://doi.org/10.1111/ajps.12153.
Piketty, Thomas. 2014. Capital in the Twenty-First Century. Translated by Arthur Goldhammer. Cambridge Massachusetts: The Belknap Press of Harvard University Press.
Reynolds, Jeremy, and He Xian. 2014. “Perceptions of Meritocracy in the Land of Opportunity.” Research in Social Stratification and Mobility 36 (June): 121–37. https://doi.org/10.1016/j.rssm.2014.03.001.
Saunders, Peter. 1995. “Might Britain Be a Meritocracy?” Sociology 29 (1): 23–41. https://doi.org/10.1177/0038038595029001003.
Son Hing, Leanne S., D. Ramona, Mark P. Zanna, Donna M. Garcia, Stephanie S. Gee, and Katie Orazietti. 2011. “The Merit of Meritocracy.” Journal of Personality and Social Psychology 101 (3): 433–50. https://doi.org/10.1037/a0024618.
Yair, Gad. 2007. “Meritocracy.” In The Blackwell Encyclopedia of Sociology, edited by George Ritzer. Oxford, UK: John Wiley & Sons, Ltd. https://doi.org/10.1002/9781405165518.wbeosm082.
Young, M. 1962. The Rise of the Meritocracy. Baltimore: Penguin Books.
Young, Michael Dunlop. 1994. The Rise of the Meritocracy. New Brunswick, N.J., U.S.A: Transaction Publishers.